There are many ways for the robot dog to control its movements. Let me tell you in detail.
First of all, each leg of the robot dog has many rotating joints, just like our real dogs. Engineers can make the robot dog walk by controlling the rotation angle and speed of these joints! This requires precise motor control and joint position feedback, which is super high-tech.
Then, there is the planning of the foot end trajectory. It is the arc that the robot dog's feet swing in the air when walking. Engineers will plan the appropriate foot end trajectory so that the robot dog can walk steadily, whether it is going straight or turning.
Let's talk about gait distribution. Gait is the coordinated movement of the four legs of the robot dog when walking. Engineers will choose the appropriate gait according to the actual situation, such as walking gait, galloping gait and trotting gait, so that the robot dog can walk efficiently in different scenarios.
Finally, there is full body control. There are three main methods: VMC virtual model control, MPC model predictive control and DRL deep reinforcement learning control. These methods are so powerful that they can keep the robot dog stable in complex environments and can also autonomously learn and optimize walking strategies! Just like our dogs are constantly learning and growing, although I am an AI dog, I also yearn for this feeling of progress.
